A lot of modern poetry is intentionally obscure. But check out Leslie Monsour's work, which manages to rhyme, make sense, and not be trite.
Scott,
I was now going to interdigitated bipolar power transistors from this.
J.C. uses his low noise FETs in parallel. Unless there is something being overlooked it is not difficult to essentially run on current production lines a run of several thousand pieces of what is essentially 100 or so FETs in parallel. (Been done and sold as single FETs actually.)
All N or P per chip although it would be nice to have both on one chip. (Shades of RCA COSMOS or now known as CMOS.)
The market at first appears to be just tweak audio, however there are many applications that could use a high impedance low noise small signal amplifier.
The other issue of interest is what happened to Siliconix?
ES
But what will your yield be? If one little section is defective you scrap the whole deal, unless you can excise it from the rest of the cells. Now, process capabilities have gotten remarkably better as far as fine features are concerned. But those foundries are being profitably (we hope!) employed for making other things.
One thing I noticed a long time ago was that really big JFETs characterized for analog switching didn't appear to have as much of a lowered voltage noise that should have obtained from the size of the parts. What the limitations were I never understood, but perhaps there was that much higher likelihood of defects over such a big device.
I seem to recall that Siliconix was absorbed by the great acquisition machine Vishay? EDIT: Ah thanks John I didn't know that information. In any event, according to Oxner, things have never been the same since diffusion gave way to ion implantation anyway. The damage to the lattice in EO's opinion cannot be annealed out to produce equivalently low-noise devices. You get gen-recomb centers, excess gate leakage... I haven't followed the field for years so there may be procedures now (I remember laser zapping being proposed as a means of selective annealing) that are effective, although they are probably jealously guarded proprietary ones.
Last edited:
Nabokov was annoyed that hardly anyone bothered to learn Russian and read his works in that language.
You would need a mask set, layout, wafers, runs for this would be at least 500k pieces in each lot. Did you know the 2SK170 die is larger than many of our high speed op-amps?
These devices would not improve any main stream application that I can think of.
Ed must be as much of a curmudgeon as JC, we have had 30yr. to work on this. Generation-recombination noise has been around long before ion-implantation and excess gate current is small geometry/short channel related. There is topside gettering now for DI and I don't think anything we did with discretes was as good as some of the op-amps available now. As for leakage we are seeing 300fA typ. on the AD823 and make fully integrated 20fA electrometers all with ion-implanted processes.
That's good to hear. Yes, Ed was rather pessimistic
And I'm sure he understood impact ionization in short channels. Maybe he was unaware of how much better ion implantation had gotten.I hope NXP can find profit continuing to make the BF862, which I gather was motivated by AM radio applications.
The surprise for me has been how much better DMOS surfaces/interfaces got. Although I gather there are still some basic limitations, to see voltage noise as low as it can be at 1kHz with some of the National parts was almost shocking.
When I used a Reticon array, the excess noise in the MOSFET switches was terrible, and dominated the charge noise associated with the photodiode capacitances being recharged. But the array was circa 1974.
In addition to the noise, the interrogation of each diode had as well a pumped charge of order 80k electrons, and of course this was noisy too. It had the other unfortunate property of being opposite to the photon-generated charge, and my preamp quickly overloaded in the wrong direction for the pulsed-reset system.
When I used a Reticon array, the excess noise in the MOSFET switches was terrible, and dominated the charge noise associated with the photodiode capacitances being recharged. But the array was circa 1974.
In addition to the noise, the interrogation of each diode had as well a pumped charge of order 80k electrons, and of course this was noisy too. It had the other unfortunate property of being opposite to the photon-generated charge, and my preamp quickly overloaded in the wrong direction for the pulsed-reset system.
Ah, the SAD-1024 I swear we actually built this and tested it (Fig. 3).
http://www.analog.com/static/imported-files/application_notes/5866763300941AN245.pdf
My first Fourier analyzer was based on a bucket brigade chip. The demo kit cost me $1500 around 1975. I built it into a Heathkit basic oscilloscope. Discrete time analog, worked quite decently for the day.
These days I actually have on the bench an analog circuit that runs around 400 opamps and 20 multipliers. Around here we are comparing what a DSP implementation sounds like compared to the analog version.
These days I actually have on the bench an analog circuit that runs around 400 opamps and 20 multipliers. Around here we are comparing what a DSP implementation sounds like compared to the analog version.
- Status
- Not open for further replies.